Embed Notice

I'm so confused when people try to get me to invest in their generative AI company. Like, the entire concept of generative AI does not work on the most basic fundamental level. The whole appeal is that it's cheaper than paying an actual human to do it. But, to even be mediocre at any given task the model must be train on millions of data points from people who are actually good at said task. It works (temporarily) because the internet has been free and open for all of its existence, providing these companies with infinite training data at zero cost. But why would researchers and creators continue publishing their work on the open internet under a licensing model that allows someone else to profit while directly undermining their career? ChatGPT has already started to degrade as a result of being given internet access. What do we think happens when all the high quality data points start getting paywalled or licensed out of reach?